基于时域太赫兹光谱技术的橄榄油氧化程度检测研究

Detection of oxidation degree of olive oil based on time-domain terahertz spectroscopy*

文章通过设定特定储藏条件加速橄榄油氧化酸败,获得不同氧化程度的橄榄油样品,并对样品的脂肪酸变化进行分析。采用太赫兹系统对不同储藏期的橄榄油样品进行检测,选择0.3THz~2.0 THz范围内的THz吸收光谱作为建模数据,结合支持向量机(SVM)、反向传播神经网络(BPNN)、偏最小二乘(PLS)、随机森林(RF)建模方法对橄榄油储藏时间进行预测,结果表明支持向量机所得预测结果最优,再建模集中的预测正确率为98.5%,预测集中的建模正确率为99%。实验结果表明,太赫兹光谱技术结合机器学习算法可以有效检测橄榄油的氧化程度,为橄榄油的快速检测提供了新方法。

Article based on the oxidation characteristics of extra virgin olive oil(EVOO),the extraction of olive oil spectra with different degrees of oxidation was achieved.Different kinds of fat acid were analyzed during the oxidation process of the EVOO.The discriminant model was established by screening the characteristic spectra using the time-domain terahertz spectroscopy combined with support vector machines(SVM),back propagation neural network(BPNN),partial least squares(PLS),and random forest(RF).Compared with the accuracy of different models,the identification accuracy of calibration set and prediction set were 98.5%and 99%using SVM model,respectively.The results showed that terahertz spectroscopy could quickly detect the oxidation degree of olive oil and meet the real-time monitoring of the quality during processing and storage of olive oil.